LORM: Learning to Optimize for Resource Management in Wireless Networks With Few Training Samples
Open accessThe DOI matches its Crossref record; 39 of the 39 references it cites were found in a registry or an index.
Abstract
Effective resource management plays a pivotal role in wireless networks, which, unfortunately, typically results in challenging mixed-integer nonlinear programming (MINLP) problems. Machine learning-based methods have recently emerged as a disruptive way to obtain near-optimal performance for MINLPs with affordable computational complexity. There have been some attempts in applying such methods to resource management in wireless networks, but these attempts require huge amounts of training samples and lack the capability to handle constrained problems. Furthermore, they suffer from severe performance deterioration when the network parameters change, which commonly happens and is referred to as thetask mismatchproblem. In this paper, to reduce the sample complexity and address the feasibility issue, we propose a framework of Learning to Optimize for Resource Management (LORM). In contrast to the end-to-end learning approach adopted in previous studies, LORM learns the optimal pruning policy in the branch-and-bound algorithm for MINLPs via a sample-efficient method, namely,imitation learning. To further address the task mismatch problem, we develop a transfer learning method via self-imitation in LORM, namedLORM-TL, which can quickly adapt a pre-trained machine learning model to the new task with only a few additionalunlabeledtraining samples. Numerical simulations demonstrate that LORM outperforms specialized state-of-the-art algorithms and achieves near-optimal performance, w…
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“This learning based brand-and-bound method can significantly save the time for pruning the nodes in the search tree, achieve near-optimal performance with few training samples, as well as guarantee feasibility of constraints without performance degradation.”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [22]
In the citing paper
“To solve the large-scale mixed combinatorial optimization problem efficiently, imitation learning was adopted in [22] to learn an aggressive pruning policy in the globally optimal-achieving branch-and-bound algorithm.”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [22]
In the citing paper
“However, due to the distribution shift for system parameters in episodically dynamic environment, the trained model may suffer from performance deterioration when the dataset follows a different distribution in the inference stage [22].”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [22]
Citations
Cites
Pan, S. J., & Yang, Q. (2010) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2009.191 · Cited by 23,387 · cited 1×
“An essential assumption of most machine learning algorithms is that the training and future testing data are in the same feature space with the same distribution, i.e., the same task [33].”
BackgroundMastering the game of Go with deep neural networks and tree search
Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al. (2016) · Nature · doi:10.1038/nature16961 · Cited by 15,612 · cited 1×
“Modeling the tree search process as a sequential decision problem is a common method in machine learning [30].”
Method(2002) · Electronics - A First Course · doi:10.4324/9780080574714-46 · Cited by 3,877
Multiuser OFDM with adaptive subcarrier, bit, and power allocation
Wong, C. Y., Cheng, R. S., Lataief, K. B., & Murch, R. D. (1999) · IEEE Journal on Selected Areas in Communications · doi:10.1109/49.793310 · Cited by 2,657 · cited 4×
“Typical examples include subcarrier management in OFDMA [4], user association [5], network power minimization in Cloud-RANs [6], and joint task offloading scheduling and transmit power management [7].”
BackgroundInfluentialSelf-citationShi, Q., Razaviyayn, M., Luo, Z.-Q., & He, C. (2011) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2011.2147784 · Cited by 2,129 · cited 4×
“For instance, different algorithms have been developed for power control (e.g., water filling [17]), transmit beamforming design (e.g., WMMSE [18] and conic optimization-based algorithms [19]), user association (e.g., the primal dual optimization-based algorithm [5]), and dynamic offloading (e.g., the Lyapunov optimizat…”
MethodInfluentialVinyals, O., Fortunato, M., & Jaitly, N. (2015) · neural information processing systems · Cited by 2,043
A Reduction of Imitation Learning and Structured Prediction to No-Regret Online Learning
Ross, S., Gordon, G. J., & Bagnell, J. A. (2018) · Carnegie Mellon University · doi:10.1184/r1/6550949 · Cited by 1,941
Dynamic computation offloading for mobile-edge computing with energy harvesting devices
Mao, Y., Zhang, J., & Letaief, K. B. (2016) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2016.2611964 · Cited by 1,705 · cited 2×
“Typical examples include subcarrier management in OFDMA [4], user association [5], network power minimization in Cloud-RANs [6], and joint task offloading scheduling and transmit power management [7].”
BackgroundSelf-citationAn Introduction to Deep Reinforcement Learning
Vincent, F.-L., Peter, H., Riashat, I., Marc, G. B., & Joelle, P. (2018) · Foundations and Trends® in Machine Learning · doi:10.1561/2200000071 · Cited by 1,286 · cited 1×
“…found, N opt ← the node containing optimal visited nodes and features into D Save solutions of solved convex problems into T 3: if c ∗ c ∗ old then 4: N opt ← the nodes from root to N opt 5: else 6: if node ( D j ) ∈ N opt then 10: end if 14: end for 15: return c ∗ , N opt , D , T policy [34].”
BackgroundUser Association for Load Balancing in Heterogeneous Cellular Networks
Ye, Q., Rong, B., Chen, Y., Al-Shalash, M., Caramanis, C., & Andrews, J. G. (2013) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2013.040413.120676 · Cited by 1,228 · cited 3×
“, the primal dual optimization-based algorithm [5]), and dynamic offloading (e.”
BackgroundTaskonomy: Disentangling Task Transfer Learning
Zamir, A. R., Sax, A., Shen, W., Guibas, L., Malik, J., & Savarese, S. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00391 · Cited by 1,128
Learning Combinatorial Optimization Algorithms over Graphs
Dai, H., Khalil, E. B., Zhang, Y., Dilkina, B., & Song, L. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1704.01665 · Cited by 1,007
Learning to Optimize: Training Deep Neural Networks for Interference Management
Sun, H., Chen, X., Shi, Q., Hong, M., Fu, X., & Sidiropoulos, N. D. (2018) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2018.2866382 · Cited by 874 · cited 5×
“, MLP, in the test stage must be the same as in the training stage, most of the end-to-end learning methods [10]–[12] cannot generalize to settings with different problem sizes than those in the training dataset.”
BackgroundMethodInfluentialShi, Q., Razaviyayn, M., Luo, Z.-Q., & He, C. (2011) · 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp.2011.5946304 · Cited by 873
Towards a Reliable and Context-Based System Architecture for Autonomous Vehicles
Kain, T., Mundhenk, P., Müller, J.-S., Tompits, H., Wesche, M., & Decke, H. (2020) · OASIcs : OpenAccess Series in Informatics · doi:10.4230/oasics.asd.2020.1 · Cited by 797
Conic Optimization via Operator Splitting and Homogeneous Self-Dual Embedding
O’Donoghue, B., Chu, E., Parikh, N., & Boyd, S. (2016) · Journal of Optimization Theory and Applications · doi:10.1007/s10957-016-0892-3 · Cited by 664 · cited 3×
“When LORM is applied to the network power minimization problem in Cloud-RANs, which is mixed integer second order cone programming, the expected time complexity is O(L(4)K(3) +L(3)K(2) 1 ) if the conic optimization-based convex optimization solver [32] is used.”
MethodInfluentialMulti-Agent Deep Reinforcement Learning for Dynamic Power Allocation in Wireless Networks
Nasir, Y. S., & Guo, D. (2019) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2019.2933973 · Cited by 660 · cited 1×
“deep reinforcement learning [13], spatial deep learning [14], and deep neural network parameterization [15] have been”
BackgroundConforti, M., Cornuéjols, G., & Zambelli, G. (2014) · Graduate Texts in Mathematics · doi:10.1007/978-3-319-11008-0 · Cited by 546
An introduction to deep reinforcement learning
François-Lavet, V., Henderson, P., Islam, R., Bellemare, M. G., & Pineau, J. (2018) · doi:10.1561/9781680835397 · Cited by 544
Group Sparse Beamforming for Green Cloud-RAN
(2014) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2014.040214.131770 · Cited by 468 · cited 15×
“2) Iterative Group Sparse Beamforming (GSBF) [6]: This algorithm is designed for the network power minimization problem in Cloud-RANs, and is the state-of-the-art solutions for this problem.”
BackgroundMethodInfluentialSelf-citationAn algorithmic perspective on imitation learning
Osa, T., Pajarinen, J., Neumann, G., Bagnell, J. A., Abbeel, P., & Peters, J. (2018) · Foundations and Trends in Robotics · doi:10.1561/2300000053 · Cited by 405 · cited 2×
“This leads to imitation learning, a machine learning method that enjoys low sample complexity and good generalization capability [25].”
BackgroundDeep Power Control: Transmit Power Control Scheme Based on Convolutional Neural Network
Lee, W., Kim, M., & Cho, D.-H. (2018) · IEEE Communications Letters · doi:10.1109/lcomm.2018.2825444 · Cited by 305 · cited 3×
“, MLP, in the test stage must be the same as in the training stage, most of the end-to-end learning methods [10]–[12] cannot generalize to settings with different problem sizes than those in the training dataset.”
BackgroundMethodLearning to Branch in Mixed Integer Programming
Khalil, E., Le Bodic, P., Song, L., Nemhauser, G., & Dilkina, B. (2016) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v30i1.10080 · Cited by 298 · cited 1×
“(MILPs) [23], [24] and problems on graphs [25].”
BackgroundResource Allocation for Wireless Networks: Basics, Techniques, and Applications
Han, Z., & Liu, K. J. R. (2008) · Cambridge University Press eBooks · doi:10.1017/cbo9780511619748 · Cited by 295 · cited 1×
“IN WIRELESS networks, effective management of radio resources is vital for performance optimization [3].”
BackgroundLearning to Search in Branch and Bound Algorithms
He, H., Daumé, H., & Eisner, J. M. (2014) · neural information processing systems · Cited by 275
Towards Optimal Power Control via Ensembling Deep Neural Networks
Liang, F., Shen, C., Yu, W., & Wu, F. (2020) · IEEE Transactions on Communications · doi:10.1109/tcomm.2019.2957482 · Cited by 268 · cited 1×
“To achieve near-optimal performance, unsupervised learning methods have been proposed in [11], [12], which do not rely on any existing resource management algorithm.”
MethodSpatial Deep Learning for Wireless Scheduling
Cui, W., Shen, K., & Yu, W. (2019) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2019.2904352 · Cited by 265
Neural Combinatorial Optimization with Reinforcement Learning
Bello, I., Pham, H., Le, Q. V., Norouzi, M., & Bengio, S. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1611.09940 · Cited by 257
Learning Optimal Resource Allocations in Wireless Systems
Eisen, M., Zhang, C., Chamon, L. F. O., Lee, D. D., & Ribeiro, A. (2019) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2019.2908906 · Cited by 253 · cited 1×
“deep reinforcement learning [13], spatial deep learning [14], and deep neural network parameterization [15] have been”
BackgroundMachine learning for combinatorial optimization: A methodological tour d'horizon
Bengio, Y., Lodi, A., & Prouvost, A. (2021) · European Journal of Operational Research · doi:10.1016/j.ejor.2020.07.063 · Cited by 195
Zappone, A., Di Renzo, M., Debbah, M., Lam, T. T., & Qian, X. (2019) · IEEE Vehicular Technology Magazine · doi:10.1109/mvt.2019.2921627 · Cited by 190
Cheng, Y., Pesavento, M., & Philipp, A. (2013) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2013.2261993 · Cited by 125 · cited 1×
“erative algorithm. We set the number of DAgger as M= 5 and train 5 epochs for each DAgger iteration. There are three benchmarks: 1)Relaxed Mixed-integer Nonlinear Programming based algorithm (RMINLP) [39]: In this algorithm, the deflation procedure is performed to switch off RRHs one-by-one based on the solution to a r…”
MethodLarge-Scale Convex Optimization for Dense Wireless Cooperative Networks
Shi, Y., Zhang, J., O'Donoghue, B., & Letaief, K. B. (2015) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2015.2443731 · Cited by 118 · cited 1×
“, WMMSE [18] and conic optimization-based algorithms [19]), user association (e.”
MethodSelf-citationLearning to BranchPreprint
Balcan, M.-F., Dick, T., Sandholm, T., & Vitercik, E. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1803.10150 · Cited by 103
Signal Processing and Optimal Resource Allocation for the Interference Channel
Hong, M., & Luo, Z.-Q. T. (2014) · Academic press library in signal processing · doi:10.1016/b978-0-12-396500-4.00008-9 · Cited by 88
Transfer Learning for Mixed-Integer Resource Allocation Problems in Wireless Networks
Shen, Y., Shi, Y., Zhang, J., & Letaief, K. B. (2019) · International Conference on Communications · doi:10.1109/icc.2019.8761327 · Cited by 37 · cited 1×
“This article was presented in part at the IEEE Global Conference on Signal and Information Processing (GlobalSIP), 2018 [1] and the IEEE International Conference on Communications (ICC), 2019 [2].”
BackgroundSelf-citationSpatial Deep Learning for Wireless Scheduling
Cui, W., Shen, K., & Yu, W. (2018) · Globecom · doi:10.1109/glocom.2018.8647453 · Cited by 29
Zappone, A., Renzo, M. D., Debbah, M. A., Tu, L., & Qian, X. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1808.01672 · Cited by 14 · cited 1×
“This trend has also inspired researchers to apply machine learning-based methods to solve optimization problems in wireless networks [9].”
MethodSCALABLE NETWORK ADAPTATION FOR CLOUD-RANS: AN IMITATION LEARNING APPROACH
Shen, Y., Shi, Y., Zhang, J., & Letaief, K. B. (2018) · IEEE Global Conference on Signal and Information Processing · doi:10.1109/globalsip.2018.8646503 · Cited by 3 · cited 2×
“o1[1] = # optimal nodes in the training set”
BackgroundSelf-citation
Research works printed but not linked: 3
- proakis, 2001, Digital Communications, 4
- goodfellow, 2016, Deep Learning
- song, 2017, Learning to search via self-imitation with application to risk-aware planning, Proc Adv Neural Inf Process Syst Workshop
Cited by
Edge artificial intelligence for 6g: vision, enabling technologies, and applications
Letaief, K. B., Shi, Y., Lu, J., & Lu, J. (2022) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2021.3126076 · Cited by 829
“However, due to the distribution shift for system parameters in episodically dynamic environment, the trained model may suffer from performance deterioration when the dataset follows a different distribution in the inference stage [22].”
BackgroundMethodInfluentialSelf-citationNear-Field Communications: A Tutorial Review
Liu, Y., Wang, Z., Xu, J., Ouyang, C., Mu, X., & Schober, R. (2023) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2023.3305583 · Cited by 575
“Furthermore, some low-complexity algorithms, such alternating direction method of multipliers (ADMM) [72] and machine-learning-based methods [73], can be employed to obtain a high-quality sub-optimal solution.”
MethodShen, Y., Shi, Y., Zhang, J., & Letaief, K. B. (2021) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2020.3036965 · Cited by 367
“Most of existing works on “learning to optimize” approaches to solve problems in wireless networks adopted MLPs as the neural network architecture [7], [9], [11].”
BackgroundMethodInfluentialSelf-citationZhou, X., Liang, W., Shimizu, S., Ma, J., & Jin, Q. (2021) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2020.3047675 · Cited by 347
“[19] presented a machine learning-based framework for resource management in wireless communications.”
BackgroundDeep-Learning-Based Wireless Resource Allocation With Application to Vehicular Networks
Liang, L., Ye, H., Yu, G., & Li, G. Y. (2020) · Proceedings of the IEEE · doi:10.1109/jproc.2019.2957798 · Cited by 282
“Alternatively, deep learning can be embedded as a component to accelerate some steps of a well-behaved optimization algorithm, such as the pruning stage of the branch-and-bound in [39], [40].”
BackgroundMethodInfluential
Metadata
- Authors
- Shen, Y., Shi, Y., Zhang, J., & Letaief, K. B.
- Affiliations
- Hong Kong University of Science and Technology (HK)
- ShanghaiTech University (CN)
- Hong Kong Polytechnic University (HK)
- Venue
- IEEE Transactions on Wireless Communications
- Published
- 2020-01
- Volume
- 19
- Issue
- 1
- Pages
- 665-679
- ISSN
- 1536-1276 (ISSN-L)
- 1558-2248
- OpenAlex
- W2981871664
- Cited by
- 134
- References
- 37
- Citations by year
- 2026: 11 · 2025: 10 · 2024: 12 · 2023: 24 · 2022: 19 · 2021: 33 · 2020: 19
- FWCI
- 6.92
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Wireless Communications (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Rare & Special e-Zone (The Hong Kong University of Science and Technology) (submittedVersion)
- arXiv (Cornell University)
- Funders
- General Research Funding from the Research Grants Council of Hong Kong (16210719)
- National Natural Science Foundation of China (61601290)
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Wireless Network Optimization
- Physical Sciences › Computer Science › Computer Networks and Communications › Software-Defined Networks and 5G
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced MIMO Systems Optimization
- Keywords
- mixed-integer nonlinear programming, learning to optimize, imitation learning, transfer learning, self-imitation, sample efficiency, unlabeled samples, feasibility
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.